sync: update all skills from latest workspace code

doc_parser_skill:
- New: verify_flowchart.py (flowchart validation)
- Updated: LLM.py (multi-provider: DeepSeek + DashScope)
- Updated: image_parser.py (logic tree support, external prompts)
- Updated: SKILL.md, prompts/image_prompt.md

conflict_detection_skill:
- Updated: LLM.py (multi-provider sync)
- Updated: detect_conflicts.py (logic tree text conversion)

ir_generation_skill:
- Replaced old scripts/LLM.py + ir_generator.py with standalone project
- New: main.py, config.py, step1-3_*.py, ensemble_merge.py
- New: prompts/, tests/ subdirectories

tests:
- New: acceptance/ test suite with schema validation
- Fixed: conftest no longer globally skips non-acceptance tests
- Updated: test_sample.py for new ir_generation structure

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

skills/conflict_detection_skill/scripts/detect_conflicts.py

@@ -96,6 +96,77 @@ PROMPT_DETECT_CONFLICT = """你是一个文档一致性检查专家。以下内
 """
 
 
+def _is_nested_tree(lt: dict) -> bool:
+    """Return True if logic_tree uses the nested children format."""
+    return isinstance(lt.get("children"), list)
+
+
+def _logic_tree_to_text(lt: dict) -> str:
+    """Convert logic_tree JSON to readable text for conflict detection.
+
+    Supports both the new nested-tree format and the legacy flat-nodes format.
+    """
+    if _is_nested_tree(lt):
+        return _nested_tree_to_text(lt)
+    return _flat_tree_to_text(lt)
+
+
+def _nested_tree_to_text(tree: dict) -> str:
+    """Convert a nested flowchart tree to readable text."""
+    lines: list[str] = []
+
+    def _walk(node: dict, indent: int = 0):
+        prefix = "  " * indent
+        nid = node.get("id", "")
+        name = node.get("name", "")
+        ntype = node.get("type", "")
+
+        type_label = {
+            "start": "起始", "end": "结束", "process": "处理",
+            "decision": "判断", "action": "动作",
+        }.get(ntype, ntype)
+
+        lines.append(f"{prefix}[{type_label}] {nid}: {name}")
+
+        if ntype == "decision":
+            for child in node.get("children", []):
+                cond = child.get("condition", "")
+                lines.append(f"{prefix}  分支 \"{cond}\":")
+                _walk(child["node"], indent + 2)
+        elif "children" in node:
+            for child in node.get("children", []):
+                _walk(child, indent + 1)
+
+    _walk(tree)
+    return "\n".join(lines)
+
+
+def _flat_tree_to_text(lt: dict) -> str:
+    """Convert legacy flat-nodes logic_tree to readable text."""
+    lines: list[str] = []
+    root = lt.get("root", "")
+    if root:
+        lines.append(f"根节点: {root}")
+    for node in lt.get("nodes", []):
+        nid = node.get("id", "")
+        ntype = node.get("type", "")
+        if ntype == "decision":
+            cond = node.get("condition", "")
+            branches = node.get("branches", [])
+            lines.append(f"判断节点 {nid}: 条件=\"{cond}\"")
+            for b in branches:
+                lines.append(f"  - 分支 \"{b.get('value', '')}\" → {b.get('target', '')}")
+        elif ntype == "action":
+            lines.append(f"动作节点 {nid}: {node.get('description', '')}")
+        elif ntype == "state":
+            lines.append(f"状态节点 {nid}: {node.get('description', '')}")
+        elif ntype == "start":
+            lines.append(f"起始节点 {nid}: {node.get('description', '')}")
+        elif ntype == "end":
+            lines.append(f"结束节点 {nid}: {node.get('description', '')}")
+    return "\n".join(lines)
+
+
 def _build_text_for_section(sections: list[dict], section_name: str) -> str:
     """Build a single text block for the given section name."""
     texts: list[str] = []
@@ -184,8 +255,9 @@ def detect_conflicts(
         img_type = img.get("type", "other")
         rid = img.get("rid", "")
         description = img.get("description", "").strip()
+        logic_tree = img.get("logic_tree_nested") or img.get("logic_tree")
 
-        if img_type not in DIAGRAM_TYPES or not description:
+        if img_type not in DIAGRAM_TYPES or (not description and not logic_tree):
             logger.info("Skip conflict check: rid=%s type=%s", rid, img_type)
             continue
 
@@ -211,8 +283,17 @@ def detect_conflicts(
             logger.info("  [DRY RUN] would call LLM to detect conflicts")
             continue
 
+        # Enrich description with logic_tree if available
+        combined_desc = description
+        if logic_tree:
+            lt_text = _logic_tree_to_text(logic_tree)
+            if combined_desc:
+                combined_desc = f"[结构化逻辑树]\n{lt_text}\n\n[文字描述]\n{combined_desc}"
+            else:
+                combined_desc = f"[结构化逻辑树]\n{lt_text}"
+
         prompt = PROMPT_DETECT_CONFLICT.format(
-            image_description=description,
+            image_description=combined_desc,
             text_description=text_content,
             section_name=section_name,
         )